Analytica 4.0 beta Release Notes/18.104.22.168
Changes in 22.214.171.124
- Fixed major problem with CATable::GetIndexObject. Build 126.96.36.199 introduced a critical flaw in this routine, such that the CAIndex returned was in a bad state. This is a critical fix for ADE users.
- The RenderingStyle ReferenceAsText was not previously implemented, but now is.
- In the AdeTest program, when a Handle is returned for a cell value (e.g., via GetDataByLabels), and the RenderingStyle VarTermFormat is set to 1 (so that a CAObject is returned), the AdeTest program now automatically adds this to the left-hand pane of objects that are currently open. Also, if a Reference is returned, and the RenderingStyle ReferenceAsText is false, a new CATable object is added to the left-hand pane.
- CAObject::DefTable is supposed to return Nothing (nullPtr) for non-table definitions. In recent builds, this hasn't been the case, so this was fixed.
- The Font tab has been renamed to Text.
- Default colors for background fills has been changed to non-black. The defaults make the gradient/hatch example pattern more obvious, and are a more likely choice for a background fill.
- Two new built-in graph style templates were added: Blue Surround and Teal Pastel. Reminder: If you want to use one of these as the default for all graphs in your model, select it in the Style Template pulldown (you can optionally preview it from the preview tab), then select "Global Default" in the Style template pulldown. When asked, say "Yes -- The graph appearance will remain". Then end the dialog with Set Default. Using Set Default instructs Analytica to use this style for all graphs, rather than just this single graph.
- A More Colors... button has been added to the color chooser pulldown (used for selecting font or background fill colors), which brings up an advanced color picker dialog.
- The Coordinate Index has been renamed to Comparison Index.
- The pulldown labels for "Color", "Symbol" and "Symbol Size" have been renamed to "Color Key", "Symbol Key", and "Symbol Size Key".
- Several bugs in which pulldown selections did not match what was shown on the graph were fixed. These occurred only in fairly esoteric multi-D graph configurations.
Engine and Built-in Functions
- ChanceDist has been upgraded so that it can be used with named parameters, can be used inside the Random function, and supports the Over parameter.
There is a small change functionality here when the optional index parameter is omitted. (Note: It is highly recommended to include the index parameter whenever possible, since this ensures the call is array abstractable). In ChanceDist(P,A), when A is 1-D and P is multi-D but does not share any dimension with A, Analytica would previously use A's only dimension as the default parameter, but now signals an error about the ambiguity.
- Removed a false cyclic-dependency error message that could occur when a zero offset appears in Dynamic, such as in: Dynamic(0, A[Time-0]). If there is a real offset elsewhere in the loop, this might not constitute an actual cyclic dependency, so the error was spurious.
- ReadTextFile transparently handles new-line characters in all three common formats: PC, Unix and Mac. In all cases, new-lines in the result of ReadTextFile are always Chr(10), regardless of the newline standard in the underlying file. Previously this was true for PC and Unix formats, but the Mac format (newlines of [CR]) were not auto-translated.
- CumDist and ProbDist tolerate a case that Analytica 3.1 tolerated but previous Analytica 4.0 signalled an error on. This is a case where the optional index parameter is omitted (not a recommended usage), in which one of the parameters is a scalar and the other is 1-D. For example, CumDist(0.2,A). Analytica will recognize the only dimension as being the omitted third index, as 3.1 did, rather than signaling an error that the index cannot be inferred since no dimension is shared between the two parameters.
- For variables defined as distributions whose definition displays as a button in the the attribute pane, the display of the parameters has been altered slightly. They now display as written, rather than being "regenerated". Thus, the names of named parameters are shown, where they weren't before.
- If you fill in the Over parameter is a distribution function from the function finder, it now pastes as a named parameter without brackets. E.g.: Normal(0,1,Over:I,J), rather than Normal(0,1,[I,J]). Also, any parameter in any function, including UDFs, coming after an ellipsis or hidden parameter is pasted to the definition in the named syntax (which it has to be). Ellipsis parameters paste without brackets.
Saving and Loading
- Fixed a problem that could cause models saved in XML format to be corrupted.
Example Models and Libraries
- Fixed a corruption with the "Big Mac Attack.ana" example model. (the XML problem listed above).
- Added ParseCsvText to "Flat File Library.ana"
- ReadCsvFile (and the new ParseCsvText) handle Mac-style newlines correctly. Thus lines ending in [CR], [LF], or [CRLF] are all okay, while previous Csv files from Macs, with lines ending in [CR] alone, weren't parsed correctly unless the optional linesep parameter was explicitly specified as Chr(13).